[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <20220115122650.128182-1-krzysztof.kozlowski@canonical.com>
Date: Sat, 15 Jan 2022 13:26:43 +0100
From: Krzysztof Kozlowski <krzysztof.kozlowski@...onical.com>
To: Krzysztof Kozlowski <krzysztof.kozlowski@...onical.com>,
"David S. Miller" <davem@...emloft.net>,
Jakub Kicinski <kuba@...nel.org>, linux-nfc@...ts.01.org,
netdev@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: [PATCH 0/7] nfc: llcp: fix and improvements
Hi,
Patch #1:
=========
Syzbot reported an easily reproducible NULL pointer dereference which I was
struggling to analyze:
https://syzkaller.appspot.com/bug?extid=7f23bcddf626e0593a39
Although direct fix is obvious, I could not actually find the exact race
condition scenario leading to it. The patch fixes the issue - at least under
my QEMU - however all this code looks racy, so I have a feeling I am plumbing
one leak without fixing root cause.
Therefore I would appreciate some more thoughts on first commit.
The rest of patches:
====================
These are improvements, rebased on top of #1, although should be independent.
They do not fix any experienced issue, just look correct to me from the code
point of view.
Testing
=======
Under QEMU only. The NFC/LLCP code was not really tested on a device.
Best regards,
Krzysztof
Krzysztof Kozlowski (7):
nfc: llcp: fix NULL error pointer dereference on sendmsg() after
failed bind()
nfc: llcp: nullify llcp_sock->dev on connect() error paths
nfc: llcp: simplify llcp_sock_connect() error paths
nfc: llcp: use centralized exiting of bind on errors
nfc: llcp: use test_bit()
nfc: llcp: protect nfc_llcp_sock_unlink() calls
nfc: llcp: Revert "NFC: Keep socket alive until the DISC PDU is
actually sent"
net/nfc/llcp.h | 1 -
net/nfc/llcp_core.c | 9 +-------
net/nfc/llcp_sock.c | 54 ++++++++++++++++++++++++---------------------
3 files changed, 30 insertions(+), 34 deletions(-)
--
2.32.0
Powered by blists - more mailing lists